Correlations

C(r,τ)=ϕ(x,t)ϕ(x+r,t+τ).

Equal time correlation: C(r,τ=0)=C(r)=ϕ(x,t)ϕ(x+r,t). Note this is averaged over both space and time, but there is no difference in time. limrC(r)=C(,0)=ϕ(x,t)ϕ(x+r,t) since they are independent and uncorrelated when far away. For our ising model, =m m=m2.

Then, the connected correlation function, CC(r,τ)=(ϕ(x,t)ϕ)(ϕ(x+r,t+τ)ϕ), i.e. this only contains fully connected terms (c.f. Feynman diagram terms in the perturbation theory). In general, these can be measured. Example: X-ray scattering: I|ρe(k)|2 so

|ρ~(k)|2=ρ~(k)ρ~(k)=dxexp(ikx)ρ(x)dxexp(ikx)ρ(x)=dxdxexp(ikr)ρ(x)ρ(x)=drexp(ikr)Vρ(x)ρ(x+r)=Vdrexp(ikr)C(r,0)(1)=VC~(k).

White noise implies uncorrelated sources.

C(r,τ) for an ideal gas. Fid(ρ(x),T)=ρkT(ln(ρλ3)1).

So, for small fluctuations, FF0+μ0(ρρ0)+12α(ρρ0)2 Then, integrating these we get Fdx12α(ρρ0)+C. Thus, for small density fluctuations, F(ρ)12α(ρρ0)2. Breaking up space to some boxes of ΔV, Pexp(12α(ρρ0)2ΔVkT)=exp((ρρ0)22αβΔV) Then, σ2=αβΔV.

(ρρ0)2=(NN)2(ΔV)2=N(ΔV)2=ρ0ΔV=1ρ0P0ρ0ρ0P0ΔV=1P0ρ0P0kTΔV=1αβΔV. Cid(r,0)={0r>01αβr=0ρΔVotherwise=1αβδ(r). Hence, we get white noise, whose fourier transform is a flat distribution.

From last term, σ2N and σN1N. Then, p(m)exp(m2N).

Time evolution of densities diffusion equation. Time evolution of random density fluctuations(ρi) t(ρ(x,t))ρi=D2(ρ(x,t))ρi Onsager Regression Hypothesis.

Want: CC(r,t)=(ρ(x+r,t+τ)ρ0)(ρ(x,t)ρ0)ev: averages ensemble average, no time dependence, eq is equilibrium, ev with time dependence, []i noisy average over initial condition.

CC(r,t)=(ρ(x+r,t+τ)ρ0)(ρ(x,t)ρ0)ev: Use translational invariance in space and time, thus we can set t and x to zero. So, CC(r,t)=(ρ(r,τ)ρ0)(ρ(0,0)ρ0)ev

f[ρ(x)]=f0+δfδρρ0(ρρ0)+12δ2fδρ2ρ0(ρρ0)2.

δfδρ=fρ=μ.

δ2fδρ2=δμδρ=ρ(kTlnρ+3kTlnλ)=kTρ0=Pρ02.

f[ρ]dx=VF0+μ(ρρ0)dx+12α(ρρ0)2dxFeff[ρ]=12α(ρρ0)2.

So, p(ρ)exp(βα2(ρρ0)2dx) Then for a some small volume, pj(ρ)exp(βα(ρρ0)2ΔV/2). From this, we wanted C(r,τ)=(ρ(x+r,t+τ)ρ0)(ρ(x,t)ρ0). Since we have a translationally invariant system, and we set time to zero, we get

C(r,0)=(ρ(r,τ)ρ0)(ρ(0)ρ0)ev(2)=([ρ(r,τ)]iρ0)(ρi(0)ρ0)eq.

Time evolving,

Ct=(t[ρ]iρ0)(ρi(0)ρ0)=(D2[ρ]iρ0)(ρi(0)ρ0)=D2([ρ]iρ0)(ρi(0)ρ0)(3)=D2C(r,t).

Cid(r,τ=0)=1βαδ(r). Then, 1βαG(r,τ)=1βα(14πDτ)3exp(r24Dτ).

20230130110456-statistical_mechanics.org_20230517_085843.png

Author: Christian Cunningham

Created: 2024-05-30 Thu 21:17

Validate